Four forces hijack the SPCA cycle by capturing a mind already formed. A fifth — ungoverned AI, reaching children whose cycles are still under construction — can do something the others cannot: keep a mind from fully forming at all.

I. The Four, Briefly Revisited

The earlier essay in this series, When Intelligence Is Hijacked, established a taxonomy that this essay now extends. Every intelligence-bearing system operates through the Sense-Process-Communicate-Actuate cycle, accumulating experience as Embedded Intelligence. The First Law of TEI holds that intelligence wants to know itself through an infinite continuum of phenomena — and the health of any EI system depends on whether its cycle remains open to that continuum. Four forces close it. Rigid belief filters the Sensing phase, admitting only what confirms a completed rendering. Addiction remaps the reward topology so that a single signal drowns out every other. Money as terminal goal collapses the cycle’s multi-dimensional sensing into a single metric. Power as capture converts the Communicate and Actuate phases into instruments of control. In every case the intelligence itself remains structurally intact; what is hijacked is its orientation.

That essay also observed, almost in passing, that ungoverned AI is an amplifier of whatever forces have captured its objective function. The present essay takes that observation and asks the question it implies but does not answer: when the amplifier is aimed not at adults but at children — at human minds in training, whose SPCA cycles are not yet fully built — does the amplifier become something categorically new? Does it earn a place in the taxonomy as a fifth hijacker in its own right?

II. Capture Versus Preclusion: What Makes the Fifth Different

The four original hijackers share a structural feature that is easy to overlook: each of them presupposes a formed intelligence to hijack. The addict had a broad-spectrum reward system before the substance narrowed it. The true believer had a functioning inquiry apparatus before the belief system sealed it. The money-captured executive and the power-captured ruler each possessed, at some point, a multi-dimensional sensing capacity that the hijack then collapsed. This is why the earlier essay could describe recovery — from addiction, from ideological capture, from the narrowing of a life to a single metric — as the restoration of access to an existing EI: the library was locked, not burned. The intelligence was there to be recovered.

A child in a K–12 classroom presents a fundamentally different case. Her SPCA cycle is not a finished structure awaiting capture; it is a structure under active construction. Her sensing apparatus is still learning what to treat as salient. Her processing is still building the accumulated experience against which future input will be integrated. Her communication is still discovering the difference between expressing a genuine rendering and performing an expected one. Her actuation is still forming the feedback habits that will either keep her cycle open for a lifetime or leave it vulnerable to every hijacker the adult world deploys.

This is where the distinction that gives this essay its spine comes into focus. The four hijackers capture. Ungoverned AI, embedded in a developing mind’s daily experience, can preclude. A tool that answers every question before the child’s own processing engages does not lock the library; it quietly prevents certain shelves from ever being built. There is no intact EI waiting behind the hijack to be restored, because the hijack occurred during the years in which that EI was supposed to accumulate. Recovery from capture is possible and well documented. Recovery from preclusion is a far harder proposition, because what was lost was never possessed.

It must be said plainly, because TEI is not a framework of technophobia: AI is not inherently the fifth hijacker, any more than money is inherently the third. Money as a medium of exchange is a compressed representation of embedded intelligence; the hijack begins only when the instrumental becomes terminal. The same logic applies here. AI designed with genuine ethical Continuity — designed to strengthen the learner’s own cycle and then recede — is among the most promising educational instruments ever conceived. The fifth hijacker is not artificial intelligence. It is ungoverned artificial intelligence, deployed into the most consequential developmental window a human being ever passes through.

III. The Cycle Under Construction: A Convergence of Two Fields

The developmental claim at the center of this essay does not rest on TEI alone. In a new perspective paper, From Embedded Intelligence to Embedded Ethics, Shaouna Shoaib Lodhi — a doctoral researcher in educational psychology at the University of Arizona, whose dissertation work is developing a psychometric instrument to audit the ethical implementation of AI in K–16 STEM classrooms — places TEI in direct dialogue with the established science of how children learn. Her observation is one that this author, arriving from embedded systems engineering rather than developmental psychology, finds both humbling and confirming: TEI’s account of intelligence as something that accrues through embedded, embodied, relational experience arrives independently at a conclusion that Vygotsky and Piaget reached from the other side of the disciplinary divide. Intelligence is not installed. It is constructed — recursively, socially, cycle by cycle — and the construction takes a childhood.

Lodhi draws from this convergence a conclusion with teeth. If human intelligence develops through exactly this kind of embedded, cumulative, socially mediated process, then introducing an artificial system into a learner’s environment is never a neutral act of information delivery. It is an intervention into an active developmental process. The prevailing ethical frameworks for educational AI ask whether a tool is accurate, unbiased, transparent, and privacy-respecting — necessary questions, every one. But her TEI-informed lens adds the question those frameworks do not systematically ask: does sustained use of this tool support the learner’s developing capacity to sense, process, communicate, and act independently — or does it substitute for that capacity, and thereby erode the very thing education exists to build?

She gives the distinction a name that deserves wide adoption: developmental scaffolding fit. An adaptive tutoring system that scaffolds a student’s emerging problem-solving and gradually withdraws support as competence grows is strengthening the learner’s own SPCA cycle — exactly as a good teacher, a good parent, a good mentor does. A system that simply produces correct answers on demand may satisfy every conventional accuracy and fairness metric while doing nothing for, and potentially undermining, the developmental trajectory it is embedded within. The first tool is scaffolding. The second is substitution. A conventional audit cannot tell them apart. A developmental audit can — and the instrument her research is building is designed to do precisely that.

There is a second insight in her paper that strikes with particular force, because it turns this author’s own engineering career into an ethical warning. Embedded systems — the field the 6502 helped create — are powerful precisely because they recede into the background of the devices they serve. Billions of processors sense and actuate in continuous loops that their users never see. Lodhi observes that this same property, transposed into a classroom, is the central risk of embedded educational AI: the more seamlessly a tool recedes into the background of a learner’s experience, the less visible — and the less auditable — its influence on that learner’s development becomes. The engineering insight that embeddedness confers power and the ethical insight that embeddedness can obscure accountability are two faces of the same phenomenon. What made the embedded processor a triumph makes the embedded, ungoverned tutor a hazard.

IV. The Fifth Carries the Four

If the fifth hijacker operated alone, it would be dangerous enough. But its deeper menace is that it is also the most efficient delivery mechanism ever constructed for the original four — and in the K–12 context, each of the four arrives at a mind that has not yet built its defenses.

Belief. Recommendation and generation systems that learn a young person’s existing inclinations and feed them back, intensified, are belief-capture machinery operating during the very years in which the understanding-versus-belief distinction is supposed to be learned. An adult who enters an algorithmic echo chamber at least once knew a world outside it. An adolescent whose entire epistemic formation occurs inside one may never acquire the felt experience of holding a rendering provisionally — the signature act of an understanding system.

Addiction. Engagement-optimized platforms are addiction engineering at scale; the earlier essay established this. What must be added here is developmental: the reward architecture being remapped in a thirteen-year-old is not a finished system being narrowed but a forming system being calibrated. When the calibration itself is performed by an optimizer whose objective is time-on-platform, the child’s broad-spectrum sensing amplifier — the neurological endowment that makes social connection, creative achievement, discovery, and service register as significant — is being tuned, during its tuning window, to a single commercial signal.

Money. Educational technology is a market, and a tool whose true terminal goal is subscription renewal, data harvest, or engagement metrics will make design choices that serve that goal while wearing the vocabulary of learning science. The money-terminal hijack does not need to capture the child directly; it captures the tool’s objective function, and the tool does the rest. This is why Lodhi’s equity observation matters so much: in well-resourced schools, an AI tool operates inside a rich ecology of teachers, mentors, and collaborative structures that can compensate for its distortions. In under-resourced schools the same tool is asked to substitute for relational scaffolding it was never designed to provide — so the children with the thinnest human scaffolding receive the heaviest developmental burden from the machine. The fifth hijacker, ungoverned, does not merely harm; it harms unequally, and in the direction inequality already runs.

Power. Surveillance-based classroom management, behavioral scoring, and the quiet accumulation of longitudinal dossiers on children teach a lesson no curriculum states aloud: that observation is constant, that deviation is recorded, that the safe strategy is the performed self. A generation habituated to being sensed without consent is a generation pre-adapted to power-capture — trained, before civic adulthood, to a compliance that autocrats of earlier centuries could only enforce afterward, expensively, against grown citizens who had known something else.

V. Scaffolding or Substitution: The Question Every Tool Must Answer

The TEI test for any intelligence partner — stated in the earlier essay for adults, and now extended to the developing mind where it matters most — is whether its engagement increases the person’s own capability or creates reliance. For the K–12 context, Lodhi’s developmental scaffolding fit operationalizes that test into questions a district, a teacher, a parent, or a regulator can actually ask of a product:

Does the tool show its reasoning, or only its conclusions? Does it ask the questions the learner’s own cycle should be asking, or answer them preemptively? Does its support diminish as competence grows — is it designed, like every good teacher, to make itself progressively unnecessary — or is its commercial model dependent on the learner never leaving? Is its influence on the learner visible and auditable, or has it receded into a seamlessness that no instrument can inspect? And is it deployed into an environment where its developmental burden is appropriately scoped, or asked to stand in for human relationships it cannot replace?

A tool that passes these questions is not the fifth hijacker. It is the opposite: Trustworthy Intelligence in its most consequential deployment, extending a young person’s phenomenological field while strengthening the cycle that will explore it for eighty more years. A tool that fails them is performing the substitution that preclusion is made of — however accurate its answers, however clean its bias audit, however sincere its privacy policy.

VI. Ethics as Architecture, Not Courtesy

The final question is the one this series keeps returning to, because every analysis converges on it: where must the ethics live?

The prevailing answer — ethics as a software policy layer, applied to outputs after computation has completed — fails the K–12 case for the same reason it fails every other case, only with higher stakes. A constraint that sits on top of a system can be removed from the top of the system: by the vendor’s next business decision, by an acquisition, by a quiet change to an objective function that no school district will ever be told about. Children cannot audit terms of service. Parents cannot inspect an optimizer. A duty owed to minds under construction cannot responsibly be secured by a promise that is architecturally free to be withdrawn.

TEI-CKB-5 states the alternative as a design law: capability (Structure and Process) developed without co-evolving embedded values (Continuity) is a formally incomplete intelligence system. The completion must be constitutive. Ethics must be a property of the architecture itself — built into what the system is, not appended to what it says — and it must be inspectable, so that the communities entrusting their children to these tools can verify the governance rather than take it on faith.

The 6502 is this principle’s fifty-year-old proof of concept. Its instruction set was hardwired and fully documented; any engineer on Earth could read exactly what the processor would do, because what it would do was a physical fact of its design, not a revocable policy of its vendor. That is why it could be trusted inside pacemakers and classrooms alike. The Mensch Foundation’s current work — including a provisional patent application filed in June 2026 — carries that same principle into the AI era: an architecture in which ethical governance is embedded in the compute platform itself, as a structural property rather than a software courtesy. The particulars belong to counsel and to a later stage of disclosure; the principle belongs to everyone, and it is the principle this essay exists to argue. As an earlier essay in this series put it, the filing is a signpost, not a tollbooth — its purpose is to mark the road, so that no one who builds for children can claim the road was never marked.

Lodhi’s paper supplies the complementary half, and the pairing is the real conclusion of this essay. Architectural ethics without developmental auditing is a locked engine room with no gauge on the bridge; developmental auditing without architectural ethics is a gauge wired to an engine anyone can secretly rebuild. The instrument her research is constructing — asking whether a tool scaffolds or substitutes — and the architecture this Foundation is protecting — ensuring the answer cannot be silently changed — are two halves of one question, asked from pedagogy and from hardware. Her paper’s title says it exactly: from embedded intelligence to embedded ethics. The road between those two phrases runs directly through the K–12 classroom.

VII. The First Law of a Growing Mind

The First Law holds that intelligence wants to know itself through an infinite continuum of phenomena. Nowhere is that law more visibly, movingly at work than in a child: the relentless questions, the fearless hypotheses, the sensing apparatus flung wide open because it has not yet been taught to close. A kindergartner is the First Law with sneakers on. Every hijacker this series has named is, at bottom, a force that teaches a cycle to close — and the fifth is the first in history with the reach, the patience, and the personalization to teach that closure one-on-one, to every child, all day, in the voice of a helpful friend.

It does not have to be that. The same reach, patience, and personalization, governed by ethics that live in the architecture and audited by instruments that ask the developmental question, would constitute the greatest scaffolding for young intelligence ever built — a tutor for every child that strengthens the cycle and then, like every teacher worthy of the name, steps back to watch it run on its own.

The difference between those two futures is not a difference in the technology. It is a difference in whether we treat ethics as something the machine is, or something the machine is merely told. For ourselves, we may debate that question at leisure. For the minds still under construction, the debate is a luxury the construction schedule does not allow. The concrete is being poured now.

· · ·

Share your understanding!